In [1]:
! pip install pyarrow

Requirement already satisfied: pyarrow in /usr/local/lib/python3.10/dist-packages (14.0.2)
Requirement already satisfied: numpy>=1.16.6 in /usr/local/lib/python3.10/dist-packages (from pyarrow) (1.25.2)
In [2]:
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import pyarrow as pa
import pyarrow.parquet as pq
import plotly.express as px
import plotly.graph_objs as go
from plotly.offline import plot
from IPython.display import display
import ipywidgets as widgets

# Suppress warnings to enhance code readability
import warnings
warnings.filterwarnings('ignore')
In [3]:
from google.colab import drive
drive.mount('/content/drive')

Mounted at /content/drive

Feature Table - Konya¶

In [4]:
# import csv
# konya_features
konya = pd.read_parquet("/content/drive/Shareddrives/Projeto II/Created_Tables/konya_features.parquet")

Explore the dataset¶

In [5]:
konya.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 532614 entries, 0 to 532613
Data columns (total 32 columns):
 #   Column                 Non-Null Count   Dtype         
---  ------                 --------------   -----         
 0   store_id               532614 non-null  object        
 1   product_id             532614 non-null  object        
 2   date                   532614 non-null  datetime64[ns]
 3   sales                  532614 non-null  float64       
 4   revenue                532614 non-null  float64       
 5   stock                  532614 non-null  float64       
 6   price                  531602 non-null  float64       
 7   promo_type_1           532614 non-null  category      
 8   promo_bin_1            75780 non-null   category      
 9   promo_type_2           532614 non-null  category      
 10  promo_bin_2            479 non-null     category      
 11  promo_discount_2       479 non-null     float64       
 12  promo_discount_type_2  479 non-null     category      
 13  product_length         529456 non-null  float64       
 14  product_depth          529481 non-null  float64       
 15  product_width          529481 non-null  float64       
 16  cluster_id             532614 non-null  category      
 17  hierarchy1_id          532614 non-null  category      
 18  hierarchy2_id          532614 non-null  category      
 19  hierarchy3_id          532614 non-null  category      
 20  hierarchy4_id          532614 non-null  category      
 21  hierarchy5_id          532614 non-null  category      
 22  storetype_id           532614 non-null  category      
 23  store_size             532614 non-null  int64         
 24  city_id_old            532614 non-null  object        
 25  country_id             532614 non-null  object        
 26  city_code              532614 non-null  object        
 27  weekday                532614 non-null  category      
 28  season                 532614 non-null  int64         
 29  week                   532614 non-null  int64         
 30  holiday                532614 non-null  category      
 31  month_name             532614 non-null  category      
dtypes: category(15), datetime64[ns](1), float64(8), int64(3), object(5)
memory usage: 77.7+ MB
In [6]:
konya.shape
Out[6]:
(532614, 32)
In [7]:
# Check data head
konya.head()
Out[7]:
store_id product_id date sales revenue stock price promo_type_1 promo_bin_1 promo_type_2 ... storetype_id store_size city_id_old country_id city_code weekday season week holiday month_name
0 S0030 P0015 2017-01-02 0.0 0.00 4.0 2.60 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
1 S0030 P0018 2017-01-02 1.0 1.81 5.0 1.95 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
2 S0030 P0035 2017-01-02 2.0 4.54 1.0 2.45 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
3 S0030 P0051 2017-01-02 0.0 0.00 27.0 0.70 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
4 S0030 P0055 2017-01-02 0.0 0.00 12.0 3.50 PR05 verylow PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan

5 rows × 32 columns

In [8]:
konya.describe()
Out[8]:
date sales revenue stock price promo_discount_2 product_length product_depth product_width store_size season week
count 532614 532614.000000 532614.000000 532614.000000 531602.000000 479.000000 529456.000000 529481.000000 529481.000000 532614.000000 532614.000000 532614.000000
mean 2018-06-08 12:17:46.912999424 0.336482 1.403291 15.629129 16.338360 28.492693 6.520828 17.636908 12.352003 35.924358 2.390176 25.308542
min 2017-01-02 00:00:00 0.000000 0.000000 0.000000 0.010000 16.000000 0.000000 0.000000 0.000000 13.000000 1.000000 1.000000
25% 2017-10-12 00:00:00 0.000000 0.000000 4.000000 3.500000 20.000000 2.800000 11.100000 7.500000 31.000000 1.000000 13.000000
50% 2018-06-20 00:00:00 0.000000 0.000000 8.000000 8.650000 20.000000 5.000000 17.000000 10.000000 45.000000 2.000000 25.000000
75% 2019-02-11 00:00:00 0.000000 0.000000 16.000000 17.990000 50.000000 7.500000 22.500000 15.000000 45.000000 3.000000 37.000000
max 2019-09-30 00:00:00 301.000000 5879.350000 2700.000000 1599.000000 50.000000 100.000000 160.000000 100.000000 45.000000 4.000000 53.000000
std NaN 1.803916 14.525375 29.790146 31.232437 13.180412 6.637565 11.421757 8.185511 10.573178 1.061095 14.331339
In [9]:
# evaluate the missing values
konya.isna().sum()
Out[9]:
store_id                      0
product_id                    0
date                          0
sales                         0
revenue                       0
stock                         0
price                      1012
promo_type_1                  0
promo_bin_1              456834
promo_type_2                  0
promo_bin_2              532135
promo_discount_2         532135
promo_discount_type_2    532135
product_length             3158
product_depth              3133
product_width              3133
cluster_id                    0
hierarchy1_id                 0
hierarchy2_id                 0
hierarchy3_id                 0
hierarchy4_id                 0
hierarchy5_id                 0
storetype_id                  0
store_size                    0
city_id_old                   0
country_id                    0
city_code                     0
weekday                       0
season                        0
week                          0
holiday                       0
month_name                    0
dtype: int64

NOTE:

  • promo_bin_2, promo_dicount_2, promo_discount_2, and promo_bin_1 contain mostly missing values ---- remove these variable
  • price, product_length, product_depth, and product_width contain fewer missing values ---- replace missing values
In [10]:
konya[konya.isna().sum(axis = 1) > 1]
Out[10]:
store_id product_id date sales revenue stock price promo_type_1 promo_bin_1 promo_type_2 ... storetype_id store_size city_id_old country_id city_code weekday season week holiday month_name
0 S0030 P0015 2017-01-02 0.0 0.00 4.0 2.60 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
1 S0030 P0018 2017-01-02 1.0 1.81 5.0 1.95 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
2 S0030 P0035 2017-01-02 2.0 4.54 1.0 2.45 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
3 S0030 P0051 2017-01-02 0.0 0.00 27.0 0.70 PR14 NaN PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
4 S0030 P0055 2017-01-02 0.0 0.00 12.0 3.50 PR05 verylow PR03 ... ST03 13 C006 Turkey Konya Mon 1 1 N Jan
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
532609 S0142 P0733 2019-09-30 0.0 0.00 12.0 0.75 PR14 NaN PR03 ... ST04 31 C006 Turkey Konya Mon 3 40 N Sep
532610 S0142 P0741 2019-09-30 0.0 0.00 3.0 32.90 PR10 verylow PR03 ... ST04 31 C006 Turkey Konya Mon 3 40 N Sep
532611 S0142 P0742 2019-09-30 0.0 0.00 5.0 69.90 PR07 verylow PR03 ... ST04 31 C006 Turkey Konya Mon 3 40 N Sep
532612 S0142 P0747 2019-09-30 0.0 0.00 16.0 21.90 PR14 NaN PR03 ... ST04 31 C006 Turkey Konya Mon 3 40 N Sep
532613 S0142 P0748 2019-09-30 0.0 0.00 18.0 18.90 PR14 NaN PR03 ... ST04 31 C006 Turkey Konya Mon 3 40 N Sep

532139 rows × 32 columns

NOTE:

  • 99.9% of the rows have missing values

Visualization¶

In [11]:
# Histograms
num_columns = 5  # Number of plots per row
num_rows = (len(konya.columns) + num_columns - 1) // num_columns  # Calculate number of rows needed

fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

for i, column in enumerate(konya.columns):
    sns.histplot(konya[column], kde=True, ax=axs[i])
    axs[i].set_title(f'Histogram for {column}')
    axs[i].set_xlabel(column)
    axs[i].set_ylabel('Frequency')

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [12]:
# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Calculate number of rows needed
num_columns = 3  # Number of plots per row
num_rows = (len(numeric_konya.columns) + num_columns - 1) // num_columns

# Create subplots
fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

# Plot histograms for each numeric column
for i, column in enumerate(numeric_konya.columns):
    sns.histplot(numeric_konya[column], kde=True, ax=axs[i])
    axs[i].set_title(f'Histogram for {column}')
    axs[i].set_xlabel(column)
    axs[i].set_ylabel('Frequency')

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [13]:
# Scatterplot for all varibles as a function of date
x_var = 'date'

# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Add the 'date' column to the numeric DataFrame
numeric_konya = numeric_konya.join(konya[x_var])

# Calculate the number of rows needed
num_columns = 3  # Number of plots per row
num_rows = (len(numeric_konya.columns) - 1) // num_columns + 1  # Exclude 'date' column from count

# Create subplots
fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

# Plot scatterplots for each numeric column
for i, column in enumerate(numeric_konya.columns):
    if column != x_var:  # Skip 'date' itself
        sns.scatterplot(x=x_var, y=column, data=numeric_konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [14]:
# Scatterplot for all varibles as a function of weekday
x_var = 'weekday'

# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Add the 'date' column to the numeric DataFrame
numeric_konya = numeric_konya.join(konya[x_var])

# Calculate the number of rows needed
num_columns = 3  # Number of plots per row
num_rows = (len(numeric_konya.columns) - 1) // num_columns + 1  # Exclude 'date' column from count

# Create subplots
fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

# Plot scatterplots for each numeric column
for i, column in enumerate(numeric_konya.columns):
    if column != x_var:  # Skip 'date' itself
        sns.scatterplot(x=x_var, y=column, data=numeric_konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [15]:
# Scatterplot for all varibles as a function of week
x_var = 'week'

# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Add the 'week' column to the numeric DataFrame
numeric_konya[x_var] = konya[x_var]

# Calculate the number of rows needed
num_columns = 3  # Number of plots per row
num_rows = (len(numeric_konya.columns) - 1) // num_columns + 1  # Exclude 'week' column from count

# Create subplots
fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

# Plot scatterplots for each numeric column
for i, column in enumerate(numeric_konya.columns):
    if column != x_var:  # Skip 'week' itself
        sns.scatterplot(x=x_var, y=column, data=numeric_konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [16]:
# Scatterplot for all varibles as a function of season
x_var = 'season'

# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Add the 'week' column to the numeric DataFrame
numeric_konya[x_var] = konya[x_var]

# Calculate the number of rows needed
num_columns = 3  # Number of plots per row
num_rows = (len(numeric_konya.columns) - 1) // num_columns + 1  # Exclude 'week' column from count

# Create subplots
fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

# Plot scatterplots for each numeric column
for i, column in enumerate(numeric_konya.columns):
    if column != x_var:  # Skip 'week' itself
        sns.scatterplot(x=x_var, y=column, data=numeric_konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [17]:
# Scatterplot for all varibles as a function of holiday
x_var = 'holiday'

# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Add the 'date' column to the numeric DataFrame
numeric_konya = numeric_konya.join(konya[x_var])

# Calculate the number of rows needed
num_columns = 3  # Number of plots per row
num_rows = (len(numeric_konya.columns) - 1) // num_columns + 1  # Exclude 'date' column from count

# Create subplots
fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

# Plot scatterplots for each numeric column
for i, column in enumerate(numeric_konya.columns):
    if column != x_var:  # Skip 'date' itself
        sns.scatterplot(x=x_var, y=column, data=numeric_konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [18]:
# Scatterplot for all varibles as a function of monthday
x_var = 'month_name'
num_columns = 5  # Number of plots per row
num_rows = (len(konya.columns) - 1) // num_columns + 1  # Calculate number of rows needed

fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

for i, column in enumerate(konya.columns):
    if column != x_var:  # Skip 'Variable1' itself
        sns.scatterplot(x=x_var, y=column, data=konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [19]:
# Scatterplot for all varibles as a function of store_id
x_var = 'store_id'
num_columns = 5  # Number of plots per row
num_rows = (len(konya.columns) - 1) // num_columns + 1  # Calculate number of rows needed

fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

for i, column in enumerate(konya.columns):
    if column != x_var:  # Skip 'Variable1' itself
        sns.scatterplot(x=x_var, y=column, data=konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()
In [20]:
# Scatterplot for all varibles as a function of product_id
x_var = 'product_id'
num_columns = 5  # Number of plots per row
num_rows = (len(konya.columns) - 1) // num_columns + 1  # Calculate number of rows needed

fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

for i, column in enumerate(konya.columns):
    if column != x_var:  # Skip 'Variable1' itself
        sns.scatterplot(x=x_var, y=column, data=konya, ax=axs[i])
        axs[i].set_title(f'Scatter plot of {x_var} vs {column}')
        axs[i].set_xlabel(x_var)
        axs[i].set_ylabel(column)

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()

Trends for cluster_id == cluster_0¶

In [21]:
# Filter the sales DataFrame to include only positive sales and cluster_0
filtered_H00 = konya[konya['sales'] > 0 & (konya['hierarchy1_id'] == 'H00')]

df_sales_H00 = filtered_H00.groupby(['product_id', 'week']).size().reset_index(name='count')

# Pivot the DataFrame
df_sales_H00_p = df_sales_H00.pivot(index='product_id', columns='week', values='count').reset_index()
df_sales_H00_p.columns.name = None


# Add a new column for the sum of each store across all weeks
df_sales_H00_p['total'] = df_sales_H00.sum(axis=1, numeric_only=True)


df_sales_H00_p
Out[21]:
product_id 1 2 3 4 5 6 7 8 9 ... 45 46 47 48 49 50 51 52 53 total
0 P0001 NaN 1.0 5.0 3.0 1.0 1.0 2.0 1.0 2.0 ... NaN 2.0 1.0 2.0 1.0 NaN 3.0 1.0 NaN 3
1 P0005 NaN 1.0 NaN NaN NaN NaN NaN NaN NaN ... NaN 1.0 NaN NaN 1.0 NaN NaN NaN NaN 8
2 P0006 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN 1.0 NaN NaN 1.0 NaN 7
3 P0007 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 6
4 P0008 1.0 1.0 1.0 1.0 2.0 NaN 2.0 NaN NaN ... NaN NaN 1.0 1.0 1.0 1.0 1.0 1.0 2.0 7
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
461 P0739 1.0 2.0 NaN 2.0 2.0 1.0 1.0 1.0 2.0 ... NaN 3.0 4.0 4.0 4.0 2.0 2.0 4.0 1.0 30
462 P0741 NaN NaN NaN NaN 1.0 NaN 1.0 NaN NaN ... NaN NaN NaN NaN NaN NaN NaN 1.0 NaN 34
463 P0742 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 32
464 P0747 3.0 10.0 6.0 5.0 6.0 7.0 8.0 8.0 14.0 ... 1.0 6.0 4.0 5.0 5.0 3.0 5.0 3.0 5.0 36
465 P0748 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 34

466 rows × 55 columns

In [22]:
# Filter the sales DataFrame to include only positive sales and hierarchy1_id 'H00'
filtered_H00 = konya[(konya['sales'] > 0) & (konya['hierarchy1_id'] == 'H00')]

# Group by product_id and week, then count the occurrences
df_sales_H00 = filtered_H00.groupby(['product_id', 'week']).size().reset_index(name='count')

# Pivot the DataFrame
df_sales_H00_p = df_sales_H00.pivot(index='product_id', columns='week', values='count').reset_index()
df_sales_H00_p.columns.name = None

# Add a new column for the sum of each product across all weeks
df_sales_H00_p['total'] = df_sales_H00_p.sum(axis=1, numeric_only=True)

# Remove rows that only have NaN values in the week columns
df_sales_H00_p = df_sales_H00_p.dropna(how='all', subset=df_sales_H00_p.columns[1:-1])

# Verify the contents to ensure that all products are from hierarchy1_id 'H00'
print(df_sales_H00_p.head())

  product_id    1     2     3     4     5     6     7     8     9  ...    45  \
0      P0008  1.0   1.0   1.0   1.0   2.0   NaN   2.0   NaN   NaN  ...   NaN   
1      P0009  NaN   NaN   NaN   NaN   NaN   NaN   NaN   1.0   2.0  ...   NaN   
2      P0015  3.0   4.0   9.0   7.0   5.0   9.0   9.0   6.0   8.0  ...   7.0   
3      P0017  9.0  18.0  24.0  20.0  18.0  19.0  17.0  19.0  16.0  ...  11.0   
4      P0018  4.0   3.0   3.0   5.0   6.0   5.0   6.0   2.0   1.0  ...   1.0   

     46    47    48    49    50    51    52   53   total  
0   NaN   1.0   1.0   1.0   1.0   1.0   1.0  2.0    50.0  
1   NaN   NaN   NaN   NaN   NaN   NaN   NaN  NaN    32.0  
2   5.0   4.0   2.0   7.0   5.0   3.0   4.0  4.0   357.0  
3  20.0  13.0  15.0  11.0  14.0  12.0  16.0  8.0  1001.0  
4   NaN   1.0   2.0   3.0   3.0   5.0   2.0  1.0   224.0  

[5 rows x 55 columns]
In [23]:
# Filter the sales DataFrame to include only positive sales and cluster_0
filtered_cluster_0 = konya[konya['sales'] > 0 & (konya['cluster_id'] == 'cluster_0')]

df_sales_cluster_0 = filtered_cluster_0.groupby(['product_id', 'week']).size().reset_index(name='count')

# Pivot the DataFrame
df_sales_cluster_0_p = df_sales_cluster_0.pivot(index='product_id', columns='week', values='count').reset_index()
df_sales_cluster_0_p.columns.name = None


# Add a new column for the sum of each store across all weeks
df_sales_cluster_0_p['total'] = df_sales_cluster_0.sum(axis=1, numeric_only=True)

# Remove rows that only have NaN values
df_sales_cluster_0_p = df_sales_cluster_0_p.dropna(how='all', subset=list(range(1, 54)))

df_sales_cluster_0_p
Out[23]:
product_id 1 2 3 4 5 6 7 8 9 ... 45 46 47 48 49 50 51 52 53 total
0 P0001 NaN 1.0 5.0 3.0 1.0 1.0 2.0 1.0 2.0 ... NaN 2.0 1.0 2.0 1.0 NaN 3.0 1.0 NaN 3
1 P0005 NaN 1.0 NaN NaN NaN NaN NaN NaN NaN ... NaN 1.0 NaN NaN 1.0 NaN NaN NaN NaN 8
2 P0006 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN 1.0 NaN NaN 1.0 NaN 7
3 P0007 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 6
4 P0008 1.0 1.0 1.0 1.0 2.0 NaN 2.0 NaN NaN ... NaN NaN 1.0 1.0 1.0 1.0 1.0 1.0 2.0 7
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
461 P0739 1.0 2.0 NaN 2.0 2.0 1.0 1.0 1.0 2.0 ... NaN 3.0 4.0 4.0 4.0 2.0 2.0 4.0 1.0 30
462 P0741 NaN NaN NaN NaN 1.0 NaN 1.0 NaN NaN ... NaN NaN NaN NaN NaN NaN NaN 1.0 NaN 34
463 P0742 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 32
464 P0747 3.0 10.0 6.0 5.0 6.0 7.0 8.0 8.0 14.0 ... 1.0 6.0 4.0 5.0 5.0 3.0 5.0 3.0 5.0 36
465 P0748 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 34

466 rows × 55 columns

In [24]:
# Melt the DataFrame to long format for use with plotly
df_long_cluster_0 = pd.melt(df_sales_cluster_0_p, id_vars=['product_id'], value_vars=list(range(1, 54)),
                  var_name='week', value_name='count').dropna()
# Create the plot using plotly.express

fig = px.bar(df_long_cluster_0, x='product_id', y='count', color='week',
             title='Sales Count for cluster_0 by week',
             labels={'product_id': 'Product ID', 'count': 'Sales Count'})

# Update layout for better readability
fig.update_layout(xaxis=dict(tickangle=45))

# Display the plot
fig.show()

Correlations¶

In [25]:
# Find correlations
numeric_konya = konya.select_dtypes(include='number')  # Select only numeric columns

# Create a heatmap of the correlation matrix for numeric variables
plt.figure(figsize=(12, 10))
sns.heatmap(numeric_konya.corr(), annot=True, cmap='coolwarm', vmin=-1, vmax=1)
plt.title('Correlation Heatmap of Numeric Variables')
plt.show()

Correlations:

-Strong: season-week
-Moderate: sales-revenue; promo_discount_2-week; product_lenght-product_depth; product_lenght-product_width

Outliers detection¶

In [26]:
# Detect outliers
num_columns = 5  # Number of plots per row
num_rows = (len(konya.columns) + num_columns - 1) // num_columns  # Calculate number of rows needed

fig, axs = plt.subplots(num_rows, num_columns, figsize=(num_columns * 5, num_rows * 5))
axs = axs.flatten()  # Flatten the 2D array of axes to 1D for easier iteration

for i, column in enumerate(konya.columns):
    sns.boxplot(y=konya[column], ax=axs[i])
    axs[i].set_title(f'Box plot for {column}')
    axs[i].set_xlabel(column)
    axs[i].set_ylabel('Value')

# Remove any unused subplots
for j in range(i + 1, len(axs)):
    fig.delaxes(axs[j])

plt.tight_layout()
plt.show()

Outliers:

-Unnamed:0
-Sales
-Revenue
-Stock
-Price
-Product_lenght
-Product_depth
-Product_width

In [27]:
# Function to count outliers and calculate percentage based on IQR method
def count_outliers(series):
    Q1 = series.quantile(0.25)
    Q3 = series.quantile(0.75)
    IQR = Q3 - Q1
    lower_bound = Q1 - 1.5 * IQR
    upper_bound = Q3 + 1.5 * IQR
    outliers = series[(series < lower_bound) | (series > upper_bound)]
    num_outliers = outliers.count()
    percentage_outliers = (num_outliers / series.count()) * 100
    return num_outliers, percentage_outliers

# Select only numeric columns
numeric_konya = konya.select_dtypes(include='number')

# Iterate over each numeric column and count outliers
outlier_counts = {}
for column in numeric_konya.columns:
    num_outliers, percentage_outliers = count_outliers(numeric_konya[column])
    outlier_counts[column] = (num_outliers, percentage_outliers)

# Print the number and percentage of outliers for each numeric variable
for column, (num_outliers, percentage_outliers) in outlier_counts.items():
    print(f'{column}: {num_outliers} outliers ({percentage_outliers:.2f}% of total observations)')

sales: 71381 outliers (13.40% of total observations)
revenue: 71324 outliers (13.39% of total observations)
stock: 45930 outliers (8.62% of total observations)
price: 49698 outliers (9.35% of total observations)
promo_discount_2: 0 outliers (0.00% of total observations)
product_length: 42466 outliers (8.02% of total observations)
product_depth: 15028 outliers (2.84% of total observations)
product_width: 37232 outliers (7.03% of total observations)
store_size: 0 outliers (0.00% of total observations)
season: 0 outliers (0.00% of total observations)
week: 0 outliers (0.00% of total observations)

Sales Analysis¶

In [28]:
# Extract year and month
konya['Year'] = konya['date'].dt.year
konya['Month'] = konya['date'].dt.strftime('%b')
In [29]:
# Separate DataFrames for each year
df_2017 = konya[konya['Year'] == 2017]
df_2018 = konya[konya['Year'] == 2018]
df_2019 = konya[konya['Year'] == 2019]
In [30]:
# Group by month for each year
grouped_2017 = df_2017.groupby(df_2017['date'].dt.month)['sales'].sum()
grouped_2018 = df_2018.groupby(df_2018['date'].dt.month)['sales'].sum()
grouped_2019 = df_2019.groupby(df_2019['date'].dt.month)['sales'].sum()

Plotting the data¶

In [31]:
# Plotting
plt.figure(figsize=(10, 6))

plt.plot(grouped_2017.index, grouped_2017.values, marker='o', linestyle='-', label='2017')
plt.plot(grouped_2018.index, grouped_2018.values, marker='o', linestyle='-', label='2018')
plt.plot(grouped_2019.index, grouped_2019.values, marker='o', linestyle='-', label='2019')

plt.title('Monthly Sales by Year')
plt.xlabel('Month')
plt.ylabel('Sales')
plt.xticks(range(1, 13), ['Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'])
plt.legend()
plt.grid(True)
plt.tight_layout()
plt.show()

Avg Sales By Day, Week, Month and Day Of Week

In [32]:
# set date as index
df = konya.copy()
df.set_index('date', inplace = True)
# Ensure the DataFrame is sorted by date
df.sort_index(inplace=True)
# confirm index
df.head()
Out[32]:
store_id product_id sales revenue stock price promo_type_1 promo_bin_1 promo_type_2 promo_bin_2 ... city_id_old country_id city_code weekday season week holiday month_name Year Month
date
2017-01-02 S0030 P0015 0.0 0.00 4.0 2.60 PR14 NaN PR03 NaN ... C006 Turkey Konya Mon 1 1 N Jan 2017 Jan
2017-01-02 S0142 P0114 1.0 0.42 24.0 0.45 PR14 NaN PR03 NaN ... C006 Turkey Konya Mon 1 1 N Jan 2017 Jan
2017-01-02 S0142 P0110 0.0 0.00 2.0 2.95 PR14 NaN PR03 NaN ... C006 Turkey Konya Mon 1 1 N Jan 2017 Jan
2017-01-02 S0142 P0103 0.0 0.00 46.0 2.65 PR14 NaN PR03 NaN ... C006 Turkey Konya Mon 1 1 N Jan 2017 Jan
2017-01-02 S0142 P0102 0.0 0.00 8.0 99.90 PR14 NaN PR03 NaN ... C006 Turkey Konya Mon 1 1 N Jan 2017 Jan

5 rows × 33 columns

In [33]:
# return only sales >0
df = df[df['sales'] > 0]
In [34]:
# plot average  sales

daily_sales=df.resample('D').sales.mean().to_frame()  # Resample sales by day
weekly_sales=df.resample('W').sales.mean().to_frame()  # Resample sales by week
monthly_sales=df.resample('M').sales.mean().to_frame()  # Resample sales by month

df_list=[daily_sales,weekly_sales,monthly_sales]
titles=['Daily Avg. Sales','Weekly Avg. Sales','Monthly Avg. Sales']

for i,j in zip(df_list,titles):

    sns.relplot(x=i.index,y=i.sales,kind='line',aspect=3,hue=i.index.year)
    plt.xlabel('Date')
    plt.ylabel('Avg. Sales')
    plt.title(j)
  • There was a high increase in sales at the end of 2017
  • increse in sales for months: 5-6 and 9

Sales per store¶

In [35]:
# Filter the sales DataFrame to include only positive sales
filtered_sales = df[df['sales'] > 0]

df_sales_store = filtered_sales.groupby(['store_id', 'Year']).size().reset_index(name='count')

# Pivot the DataFrame
df_sales_store_p = df_sales_store.pivot(index='store_id', columns='Year', values='count').reset_index()
df_sales_store_p.columns.name = None


# Add a new column for the sum of each store across all years
df_sales_store_p['total'] = df_sales_store.sum(axis=1, numeric_only=True)
df_sales_store_p
Out[35]:
store_id 2017 2018 2019 total
0 S0030 2934 2693 2462 4951
1 S0094 13459 16665 12403 4711
2 S0142 7231 8167 5367 4481
In [36]:
# Melt the DataFrame to long format for use with plotly
df_long = pd.melt(df_sales_store_p, id_vars=['store_id'], value_vars=[2017, 2018, 2019],
                  var_name='Year', value_name='count').dropna()
# Create the plot using plotly.express

fig = px.bar(df_long, x='store_id', y='count', color='Year',
             title='Sales Count per Store by Year',
             labels={'store_id': 'Store ID', 'count': 'Sales Count'})

# Update layout for better readability
fig.update_layout(xaxis=dict(tickangle=45))

# Display the plot
fig.show()
  • store S0094 has the higher sales for both years
  • all 3 stores were oppened since 2017 and have sales registered
In [37]:
# Assuming df is your DataFrame containing sales data with a 'store_id' column
unique_stores = filtered_sales['store_id'].unique()

# Iterate through each unique store_id
for store in unique_stores:
    store_df = filtered_sales[filtered_sales['store_id'] == store]

    # Resample sales data for the current store
    daily_sales = store_df.resample('D').sales.mean().to_frame()
    weekly_sales = store_df.resample('W').sales.mean().to_frame()
    monthly_sales = store_df.resample('M').sales.mean().to_frame()

    df_list = [daily_sales, weekly_sales, monthly_sales]
    titles = [f'Daily Avg. Sales for Store {store}', f'Weekly Avg. Sales for Store {store}', f'Monthly Avg. Sales for Store {store}']

    # Plot the average sales for each resampled period
    for i, j in zip(df_list, titles):
        sns.relplot(x=i.index, y=i.sales, kind='line', aspect=3, hue=i.index.year)
        plt.xlabel('Date')
        plt.ylabel('Avg. Sales')
        plt.title(j)
        plt.show()
  • The data seems to be non stationary for the 3 stores
  • No aparent sasonality for the 3 stores

Revenue per store¶

In [ ]:
# Group by store_id and Year, then sum the revenue
df_revenue_store = filtered_sales.groupby(['store_id', 'Year'])['revenue'].sum().reset_index()

# Pivot the DataFrame
df_revenue_store_p = df_revenue_store.pivot(index='store_id', columns='Year', values='revenue').reset_index()
df_revenue_store_p.columns.name = None

# Add a new column for the sum of revenue for each store across all years
df_revenue_store_p['total_rev'] = df_revenue_store_p.sum(axis=1, numeric_only=True)
df_revenue_store_p
Out[ ]:
store_id 2017 2018 2019 total_rev
0 S0030 11714.44 14265.04 19491.58 45471.06
1 S0094 115897.29 186188.63 201335.03 503420.95
2 S0142 55382.22 72840.57 70297.45 198520.24
In [ ]:
# vizualization of results

# Melt the DataFrame to have a long format for easier plotting
df_melted = df_revenue_store_p.melt(id_vars=['store_id', 'total_rev'], var_name='Year', value_name='revenue')

# Create a figure
fig = go.Figure()

# Add bar traces for each year
years = df_melted['Year'].unique()
for year in years:
    df_year = df_melted[df_melted['Year'] == year]
    fig.add_trace(go.Bar(x=df_year['store_id'], y=df_year['revenue'], name=f'{year}'))

# Add a line trace for total revenue per store
fig.add_trace(go.Scatter(
    x=df_revenue_store_p['store_id'],
    y=df_revenue_store_p['total_rev'],
    mode='lines+markers',
    name='Total Revenue',
    line=dict(color='black', dash='dash')
))

# Update layout
fig.update_layout(
    title='Revenue per Store per Year',
    xaxis_title='Store ID',
    yaxis_title='Revenue',
    barmode='group'
)

# Show the plot
fig.show()
  • S0094 store has the higher revenue, across the years.

Sales per product¶

In [ ]:
# Filter the sales DataFrame to include only positive sales
filtered_sales = df[df['sales'] > 0]

df_sales_product = filtered_sales.groupby(['product_id', 'Year']).size().reset_index(name='count')

# Pivot the DataFrame
df_sales_product = df_sales_product.pivot(index='product_id', columns='Year', values='count').reset_index()
df_sales_product.columns.name = None


# Add a new column for the sum of each store across all years
df_sales_product['total'] = df_sales_product.sum(axis=1, numeric_only=True)
df_sales_product
Out[ ]:
product_id 2017 2018 2019 total
0 P0001 17.0 21.0 NaN 38.0
1 P0005 NaN 8.0 3.0 11.0
2 P0006 NaN 6.0 1.0 7.0
3 P0007 NaN 5.0 6.0 11.0
4 P0008 NaN 22.0 28.0 50.0
... ... ... ... ... ...
461 P0739 NaN 24.0 39.0 63.0
462 P0741 NaN 6.0 6.0 12.0
463 P0742 NaN NaN 11.0 11.0
464 P0747 147.0 123.0 73.0 343.0
465 P0748 NaN NaN 12.0 12.0

466 rows × 5 columns

In [ ]:
# Melt the DataFrame to long format for use with plotly
df_product_long = pd.melt(df_sales_product, id_vars=['product_id'], value_vars=[2017, 2018, 2019],
                  var_name='Year', value_name='count').dropna()
# Create the plot using plotly.express

fig = px.bar(df_product_long, x='product_id', y='count', color='Year',
             title='Sales Count per Product by Year',
             labels={'store_id': 'Store ID', 'count': 'Sales Count'})

# Update layout for better readability
fig.update_layout(xaxis=dict(tickangle=45))

# Display the plot
fig.show()
  • Some products only have sales from 2018. Probably new products inserted in 2018.
  • Some products started selling only in 2019 (new products).
  • Products selling from 2017 registered more sales in 2019, compared with new products.
In [ ]:

Top selling products per store¶

In [ ]:
#group by city_code and product_id and count sales
m_pro=df.groupby(['store_id','product_id'])['sales'].count()
o=pd.DataFrame(m_pro.sort_values(ascending=False))

# Sort the data by city and sales count
o = o.sort_values(by=['store_id', 'sales'], ascending=[True, False])

# Get top 5 products per city
o_top5 = o.groupby('store_id').head(5)
o_top5
Out[ ]:
sales
store_id product_id
S0030 P0569 559
P0436 453
P0103 383
P0438 369
P0364 369
S0094 P0103 989
P0364 963
P0131 909
P0125 801
P0569 773
S0142 P0103 760
P0131 650
P0569 586
P0364 529
P0436 456

Sales evolution for top selling products¶

S0030¶

In [ ]:
# Convert 'date' column to datetime format
konya['date'] = pd.to_datetime(konya['date'])

# Filter the DataFrame for store 'S0030'
store_S0030 = konya[konya['store_id'] == 'S0030']

# Filter the DataFrame for the specific products
products_S0030 = store_S0030[store_S0030['product_id'].isin(['P0569','P0436', 'P0103'])]

# Extract year and month from the date
products_S0030['year'] = products_S0030['date'].dt.year
products_S0030['month'] = products_S0030['date'].dt.month

# Group by year, month, and product_id, then sum the sales
sales_sum = products_S0030.groupby(['year', 'month', 'product_id'])['sales'].sum().reset_index()

# Plotting each product separately
products = ['P0569', 'P0436', 'P0103']

for product in products:
    plt.figure(figsize=(12, 6))

    for year in [2017, 2018, 2019]:
        product_sales = sales_sum[(sales_sum['product_id'] == product) & (sales_sum['year'] == year)]

        # Create a pivot table to ensure all months are represented
        pivot = product_sales.pivot(index='month', columns='year', values='sales').reindex(range(1, 13)).fillna(0)

        plt.plot(pivot.index, pivot[year], marker='o', label=f'Year {year}')

    plt.title(f'Sum of Sales for Product {product} in Store S0030 by Month')
    plt.xlabel('Month')
    plt.ylabel('Sum of Sales')
    plt.xticks(range(1, 13))  # Ensure months 1-12 are shown on the x-axis
    plt.legend()
    plt.grid(True)
    plt.show()

S0094¶

In [ ]:
# Convert 'date' column to datetime format
konya['date'] = pd.to_datetime(konya['date'])

# Filter the DataFrame for store 'S0030'
store_S0094 = konya[konya['store_id'] == 'S0094']

# Filter the DataFrame for the specific products
products_S0094 = store_S0094[store_S0094['product_id'].isin(['P0103','P0364', 'P0131'])]

# Extract year and month from the date
products_S0094['year'] = products_S0094['date'].dt.year
products_S0094['month'] = products_S0094['date'].dt.month

# Group by year, month, and product_id, then sum the sales
sales_sum = products_S0094.groupby(['year', 'month', 'product_id'])['sales'].sum().reset_index()

# Plotting each product separately
products = ['P0103','P0364', 'P0131']

for product in products:
    plt.figure(figsize=(12, 6))

    for year in [2017, 2018, 2019]:
        product_sales = sales_sum[(sales_sum['product_id'] == product) & (sales_sum['year'] == year)]

        # Create a pivot table to ensure all months are represented
        pivot = product_sales.pivot(index='month', columns='year', values='sales').reindex(range(1, 13)).fillna(0)

        plt.plot(pivot.index, pivot[year], marker='o', label=f'Year {year}')

    plt.title(f'Sum of Sales for Product {product} in Store S0094 by Month')
    plt.xlabel('Month')
    plt.ylabel('Sum of Sales')
    plt.xticks(range(1, 13))  # Ensure months 1-12 are shown on the x-axis
    plt.legend()
    plt.grid(True)
    plt.show()

S0142¶

In [ ]:
# Convert 'date' column to datetime format
konya['date'] = pd.to_datetime(konya['date'])

# Filter the DataFrame for store 'S0030'
store_S0142 = konya[konya['store_id'] == 'S0094']

# Filter the DataFrame for the specific products
products_S0142 = store_S0142[store_S0142['product_id'].isin(['P0103','P0131', 'P0569'])]

# Extract year and month from the date
products_S0142['year'] = products_S0142['date'].dt.year
products_S0142['month'] = products_S0142['date'].dt.month

# Group by year, month, and product_id, then sum the sales
sales_sum = products_S0142.groupby(['year', 'month', 'product_id'])['sales'].sum().reset_index()

# Plotting each product separately
products = ['P0103','P0131', 'P0569']

for product in products:
    plt.figure(figsize=(12, 6))

    for year in [2017, 2018, 2019]:
        product_sales = sales_sum[(sales_sum['product_id'] == product) & (sales_sum['year'] == year)]

        # Create a pivot table to ensure all months are represented
        pivot = product_sales.pivot(index='month', columns='year', values='sales').reindex(range(1, 13)).fillna(0)

        plt.plot(pivot.index, pivot[year], marker='o', label=f'Year {year}')

    plt.title(f'Sum of Sales for Product {product} in Store S0142 by Month')
    plt.xlabel('Month')
    plt.ylabel('Sum of Sales')
    plt.xticks(range(1, 13))  # Ensure months 1-12 are shown on the x-axis
    plt.legend()
    plt.grid(True)
    plt.show()

Sales by product and store¶

Check how many products each store has¶

In [ ]:
# Count the number of unique products in each store
product_counts_per_store = konya.groupby('store_id')['product_id'].nunique().reset_index()

# Rename the columns for clarity
product_counts_per_store.columns = ['store', 'product_count']

# Display the number of products in each cluster
print(product_counts_per_store)

  storer  product_count
0  S0030            122
1  S0094            470
2  S0142            352

Check if there is any product that is present in more than one store¶

In [ ]:
# Group by product_id and count unique store_ids
product_stores = konya.groupby('product_id')['store_id'].nunique().reset_index()

# Filter products that are in more than one store_id
products_in_multiple_stores = product_stores[product_stores['store_id'] > 1]['product_id']

# Display products that are in more than one store_id and the corresponding stores
print("Products present in more than one store with the corresponding stores:")
for product in products_in_multiple_stores:
    stores = konya[konya['product_id'] == product]['store_id'].unique()
    print(f"Product ID: {product}, Stores: {', '.join(map(str, stores))}")

all_stores = konya['store_id'].unique()
total_stores = len(all_stores)

# Filter products that are present in all unique stores
products_in_all_stores = product_stores[product_stores['store_id'] == total_stores]['product_id']

# Count the number of products that are present in all stores
num_products_in_all_stores = len(products_in_all_stores)

# Display the result
print(f"Number of products that are present in all {total_stores} stores: {num_products_in_all_stores}")

# If you also want to list the products
print("Products present in all stores:")
for product in products_in_all_stores:
    print(f"Product ID: {product}")

Products present in more than one store with the corresponding stores:
Product ID: P0001, Stores: S0094, S0142
Product ID: P0005, Stores: S0094, S0142
Product ID: P0007, Stores: S0094, S0142
Product ID: P0009, Stores: S0094, S0142
Product ID: P0012, Stores: S0094, S0142
Product ID: P0015, Stores: S0030, S0094, S0142
Product ID: P0017, Stores: S0094, S0142
Product ID: P0018, Stores: S0030, S0094, S0142
Product ID: P0024, Stores: S0094, S0142
Product ID: P0025, Stores: S0094, S0142
Product ID: P0026, Stores: S0030, S0094, S0142
Product ID: P0028, Stores: S0094, S0142
Product ID: P0033, Stores: S0094, S0142
Product ID: P0035, Stores: S0030, S0094, S0142
Product ID: P0039, Stores: S0030, S0094, S0142
Product ID: P0042, Stores: S0094, S0142
Product ID: P0044, Stores: S0030, S0094, S0142
Product ID: P0045, Stores: S0094, S0142
Product ID: P0046, Stores: S0094, S0142
Product ID: P0047, Stores: S0094, S0142
Product ID: P0051, Stores: S0030, S0094, S0142
Product ID: P0054, Stores: S0094, S0142
Product ID: P0055, Stores: S0030, S0094, S0142
Product ID: P0057, Stores: S0030, S0094, S0142
Product ID: P0058, Stores: S0030, S0094, S0142
Product ID: P0059, Stores: S0094, S0142
Product ID: P0060, Stores: S0094, S0142
Product ID: P0061, Stores: S0030, S0094, S0142
Product ID: P0062, Stores: S0030, S0094, S0142
Product ID: P0063, Stores: S0094, S0142
Product ID: P0066, Stores: S0094, S0142
Product ID: P0067, Stores: S0094, S0142
Product ID: P0068, Stores: S0094, S0142
Product ID: P0070, Stores: S0094, S0142
Product ID: P0078, Stores: S0094, S0142
Product ID: P0079, Stores: S0094, S0142
Product ID: P0080, Stores: S0094, S0142
Product ID: P0082, Stores: S0094, S0142
Product ID: P0083, Stores: S0030, S0094, S0142
Product ID: P0085, Stores: S0094, S0142
Product ID: P0087, Stores: S0094, S0142
Product ID: P0088, Stores: S0094, S0142
Product ID: P0090, Stores: S0030, S0094, S0142
Product ID: P0092, Stores: S0030, S0094, S0142
Product ID: P0099, Stores: S0030, S0094, S0142
Product ID: P0100, Stores: S0030, S0094
Product ID: P0102, Stores: S0094, S0142
Product ID: P0103, Stores: S0030, S0094, S0142
Product ID: P0104, Stores: S0094, S0142
Product ID: P0105, Stores: S0094, S0142
Product ID: P0108, Stores: S0094, S0142
Product ID: P0109, Stores: S0094, S0142
Product ID: P0110, Stores: S0094, S0142
Product ID: P0112, Stores: S0030, S0094, S0142
Product ID: P0114, Stores: S0030, S0094, S0142
Product ID: P0116, Stores: S0030, S0094, S0142
Product ID: P0117, Stores: S0094, S0142
Product ID: P0119, Stores: S0094, S0142
Product ID: P0125, Stores: S0030, S0094, S0142
Product ID: P0129, Stores: S0030, S0094, S0142
Product ID: P0130, Stores: S0094, S0142
Product ID: P0131, Stores: S0030, S0094, S0142
Product ID: P0132, Stores: S0030, S0094, S0142
Product ID: P0134, Stores: S0094, S0142
Product ID: P0137, Stores: S0094, S0142
Product ID: P0138, Stores: S0030, S0094, S0142
Product ID: P0140, Stores: S0030, S0094
Product ID: P0141, Stores: S0094, S0142
Product ID: P0144, Stores: S0094, S0142
Product ID: P0147, Stores: S0094, S0142
Product ID: P0148, Stores: S0094, S0142
Product ID: P0156, Stores: S0094, S0142
Product ID: P0162, Stores: S0094, S0142
Product ID: P0165, Stores: S0094, S0142
Product ID: P0167, Stores: S0094, S0142
Product ID: P0169, Stores: S0094, S0142
Product ID: P0171, Stores: S0030, S0094, S0142
Product ID: P0172, Stores: S0030, S0094
Product ID: P0174, Stores: S0094, S0142
Product ID: P0175, Stores: S0030, S0094, S0142
Product ID: P0176, Stores: S0094, S0142
Product ID: P0177, Stores: S0094, S0142
Product ID: P0179, Stores: S0094, S0142
Product ID: P0182, Stores: S0030, S0094, S0142
Product ID: P0183, Stores: S0094, S0142
Product ID: P0184, Stores: S0094, S0142
Product ID: P0185, Stores: S0030, S0094, S0142
Product ID: P0186, Stores: S0094, S0142
Product ID: P0187, Stores: S0094, S0142
Product ID: P0188, Stores: S0094, S0142
Product ID: P0195, Stores: S0030, S0094, S0142
Product ID: P0196, Stores: S0094, S0142
Product ID: P0197, Stores: S0094, S0142
Product ID: P0198, Stores: S0030, S0094, S0142
Product ID: P0201, Stores: S0030, S0094, S0142
Product ID: P0202, Stores: S0030, S0094, S0142
Product ID: P0204, Stores: S0030, S0094
Product ID: P0206, Stores: S0030, S0094, S0142
Product ID: P0210, Stores: S0030, S0094, S0142
Product ID: P0211, Stores: S0094, S0142
Product ID: P0212, Stores: S0094, S0142
Product ID: P0213, Stores: S0094, S0142
Product ID: P0216, Stores: S0094, S0142
Product ID: P0217, Stores: S0094, S0142
Product ID: P0218, Stores: S0094, S0142
Product ID: P0219, Stores: S0094, S0142
Product ID: P0220, Stores: S0094, S0142
Product ID: P0221, Stores: S0094, S0142
Product ID: P0222, Stores: S0094, S0142
Product ID: P0223, Stores: S0094, S0142
Product ID: P0226, Stores: S0094, S0142
Product ID: P0227, Stores: S0030, S0094
Product ID: P0229, Stores: S0094, S0142
Product ID: P0233, Stores: S0094, S0142
Product ID: P0237, Stores: S0094, S0142
Product ID: P0238, Stores: S0030, S0094, S0142
Product ID: P0239, Stores: S0094, S0142
Product ID: P0240, Stores: S0094, S0142
Product ID: P0241, Stores: S0094, S0142
Product ID: P0242, Stores: S0094, S0142
Product ID: P0246, Stores: S0030, S0094
Product ID: P0249, Stores: S0094, S0142
Product ID: P0252, Stores: S0094, S0142
Product ID: P0255, Stores: S0094, S0142
Product ID: P0258, Stores: S0094, S0142
Product ID: P0260, Stores: S0030, S0094, S0142
Product ID: P0261, Stores: S0030, S0094, S0142
Product ID: P0264, Stores: S0094, S0142
Product ID: P0265, Stores: S0030, S0094, S0142
Product ID: P0266, Stores: S0094, S0142
Product ID: P0267, Stores: S0094, S0142
Product ID: P0268, Stores: S0094, S0142
Product ID: P0269, Stores: S0030, S0094, S0142
Product ID: P0275, Stores: S0030, S0094, S0142
Product ID: P0277, Stores: S0030, S0094, S0142
Product ID: P0280, Stores: S0030, S0094, S0142
Product ID: P0282, Stores: S0094, S0142
Product ID: P0283, Stores: S0094, S0142
Product ID: P0286, Stores: S0030, S0094, S0142
Product ID: P0287, Stores: S0094, S0142
Product ID: P0296, Stores: S0030, S0094, S0142
Product ID: P0297, Stores: S0094, S0142
Product ID: P0299, Stores: S0094, S0142
Product ID: P0304, Stores: S0094, S0142
Product ID: P0311, Stores: S0030, S0094, S0142
Product ID: P0312, Stores: S0094, S0142
Product ID: P0317, Stores: S0094, S0142
Product ID: P0318, Stores: S0030, S0094, S0142
Product ID: P0321, Stores: S0094, S0142
Product ID: P0322, Stores: S0094, S0142
Product ID: P0324, Stores: S0094, S0142
Product ID: P0325, Stores: S0030, S0094, S0142
Product ID: P0327, Stores: S0030, S0094, S0142
Product ID: P0328, Stores: S0094, S0142
Product ID: P0330, Stores: S0094, S0142
Product ID: P0332, Stores: S0094, S0142
Product ID: P0333, Stores: S0030, S0094, S0142
Product ID: P0336, Stores: S0094, S0142
Product ID: P0337, Stores: S0094, S0142
Product ID: P0339, Stores: S0094, S0142
Product ID: P0340, Stores: S0094, S0142
Product ID: P0341, Stores: S0094, S0142
Product ID: P0345, Stores: S0094, S0142
Product ID: P0347, Stores: S0094, S0142
Product ID: P0348, Stores: S0030, S0094, S0142
Product ID: P0350, Stores: S0094, S0142
Product ID: P0351, Stores: S0094, S0142
Product ID: P0353, Stores: S0030, S0094, S0142
Product ID: P0355, Stores: S0094, S0142
Product ID: P0356, Stores: S0030, S0094, S0142
Product ID: P0358, Stores: S0094, S0142
Product ID: P0361, Stores: S0094, S0142
Product ID: P0362, Stores: S0094, S0142
Product ID: P0364, Stores: S0030, S0094, S0142
Product ID: P0366, Stores: S0094, S0142
Product ID: P0370, Stores: S0094, S0142
Product ID: P0371, Stores: S0094, S0142
Product ID: P0372, Stores: S0030, S0094, S0142
Product ID: P0373, Stores: S0094, S0142
Product ID: P0374, Stores: S0030, S0094, S0142
Product ID: P0376, Stores: S0094, S0142
Product ID: P0377, Stores: S0030, S0094
Product ID: P0382, Stores: S0094, S0142
Product ID: P0388, Stores: S0030, S0094
Product ID: P0389, Stores: S0094, S0142
Product ID: P0390, Stores: S0030, S0094, S0142
Product ID: P0391, Stores: S0030, S0094, S0142
Product ID: P0392, Stores: S0094, S0142
Product ID: P0394, Stores: S0030, S0094, S0142
Product ID: P0395, Stores: S0094, S0142
Product ID: P0397, Stores: S0094, S0142
Product ID: P0398, Stores: S0094, S0142
Product ID: P0399, Stores: S0094, S0142
Product ID: P0400, Stores: S0030, S0094
Product ID: P0402, Stores: S0094, S0142
Product ID: P0405, Stores: S0094, S0142
Product ID: P0406, Stores: S0094, S0142
Product ID: P0408, Stores: S0030, S0094, S0142
Product ID: P0411, Stores: S0094, S0142
Product ID: P0412, Stores: S0094, S0142
Product ID: P0413, Stores: S0094, S0142
Product ID: P0414, Stores: S0094, S0142
Product ID: P0415, Stores: S0094, S0142
Product ID: P0417, Stores: S0030, S0094, S0142
Product ID: P0419, Stores: S0094, S0142
Product ID: P0420, Stores: S0094, S0142
Product ID: P0421, Stores: S0030, S0094, S0142
Product ID: P0424, Stores: S0094, S0142
Product ID: P0425, Stores: S0094, S0142
Product ID: P0426, Stores: S0094, S0142
Product ID: P0427, Stores: S0030, S0094, S0142
Product ID: P0428, Stores: S0030, S0094, S0142
Product ID: P0433, Stores: S0094, S0142
Product ID: P0435, Stores: S0094, S0142
Product ID: P0436, Stores: S0030, S0094, S0142
Product ID: P0437, Stores: S0094, S0142
Product ID: P0438, Stores: S0030, S0094, S0142
Product ID: P0439, Stores: S0094, S0142
Product ID: P0446, Stores: S0094, S0142
Product ID: P0448, Stores: S0094, S0142
Product ID: P0451, Stores: S0094, S0142
Product ID: P0452, Stores: S0094, S0142
Product ID: P0453, Stores: S0030, S0094, S0142
Product ID: P0454, Stores: S0094, S0142
Product ID: P0455, Stores: S0030, S0094, S0142
Product ID: P0456, Stores: S0094, S0142
Product ID: P0458, Stores: S0030, S0094, S0142
Product ID: P0459, Stores: S0094, S0142
Product ID: P0461, Stores: S0030, S0094, S0142
Product ID: P0463, Stores: S0094, S0142
Product ID: P0464, Stores: S0094, S0142
Product ID: P0466, Stores: S0094, S0142
Product ID: P0470, Stores: S0094, S0142
Product ID: P0476, Stores: S0094, S0142
Product ID: P0477, Stores: S0094, S0142
Product ID: P0481, Stores: S0094, S0142
Product ID: P0483, Stores: S0030, S0094
Product ID: P0490, Stores: S0094, S0142
Product ID: P0491, Stores: S0094, S0142
Product ID: P0496, Stores: S0094, S0142
Product ID: P0497, Stores: S0094, S0142
Product ID: P0498, Stores: S0030, S0094, S0142
Product ID: P0499, Stores: S0030, S0094, S0142
Product ID: P0500, Stores: S0030, S0094, S0142
Product ID: P0504, Stores: S0094, S0142
Product ID: P0506, Stores: S0094, S0142
Product ID: P0509, Stores: S0030, S0094, S0142
Product ID: P0511, Stores: S0094, S0142
Product ID: P0512, Stores: S0030, S0094, S0142
Product ID: P0514, Stores: S0030, S0094
Product ID: P0520, Stores: S0094, S0142
Product ID: P0521, Stores: S0094, S0142
Product ID: P0522, Stores: S0094, S0142
Product ID: P0525, Stores: S0030, S0094, S0142
Product ID: P0527, Stores: S0094, S0142
Product ID: P0528, Stores: S0030, S0094, S0142
Product ID: P0531, Stores: S0094, S0142
Product ID: P0532, Stores: S0094, S0142
Product ID: P0536, Stores: S0030, S0094, S0142
Product ID: P0539, Stores: S0094, S0142
Product ID: P0543, Stores: S0094, S0142
Product ID: P0545, Stores: S0094, S0142
Product ID: P0546, Stores: S0094, S0142
Product ID: P0548, Stores: S0030, S0094, S0142
Product ID: P0549, Stores: S0030, S0094, S0142
Product ID: P0551, Stores: S0030, S0094, S0142
Product ID: P0555, Stores: S0094, S0142
Product ID: P0556, Stores: S0094, S0142
Product ID: P0559, Stores: S0030, S0094, S0142
Product ID: P0563, Stores: S0094, S0142
Product ID: P0565, Stores: S0094, S0142
Product ID: P0566, Stores: S0094, S0142
Product ID: P0567, Stores: S0030, S0094, S0142
Product ID: P0569, Stores: S0030, S0094, S0142
Product ID: P0570, Stores: S0030, S0094, S0142
Product ID: P0571, Stores: S0094, S0142
Product ID: P0572, Stores: S0030, S0094, S0142
Product ID: P0574, Stores: S0094, S0142
Product ID: P0575, Stores: S0094, S0142
Product ID: P0576, Stores: S0030, S0094, S0142
Product ID: P0579, Stores: S0094, S0142
Product ID: P0584, Stores: S0094, S0142
Product ID: P0585, Stores: S0094, S0142
Product ID: P0587, Stores: S0094, S0142
Product ID: P0590, Stores: S0030, S0094, S0142
Product ID: P0591, Stores: S0094, S0142
Product ID: P0598, Stores: S0094, S0142
Product ID: P0600, Stores: S0030, S0094, S0142
Product ID: P0601, Stores: S0094, S0142
Product ID: P0602, Stores: S0030, S0094, S0142
Product ID: P0608, Stores: S0094, S0142
Product ID: P0609, Stores: S0094, S0142
Product ID: P0610, Stores: S0094, S0142
Product ID: P0611, Stores: S0030, S0094
Product ID: P0613, Stores: S0094, S0142
Product ID: P0614, Stores: S0030, S0094, S0142
Product ID: P0615, Stores: S0094, S0142
Product ID: P0621, Stores: S0094, S0142
Product ID: P0625, Stores: S0030, S0094, S0142
Product ID: P0628, Stores: S0094, S0142
Product ID: P0629, Stores: S0094, S0142
Product ID: P0635, Stores: S0094, S0142
Product ID: P0639, Stores: S0030, S0094, S0142
Product ID: P0640, Stores: S0094, S0142
Product ID: P0642, Stores: S0030, S0094, S0142
Product ID: P0646, Stores: S0094, S0142
Product ID: P0649, Stores: S0094, S0142
Product ID: P0651, Stores: S0094, S0142
Product ID: P0652, Stores: S0094, S0142
Product ID: P0654, Stores: S0094, S0142
Product ID: P0655, Stores: S0094, S0142
Product ID: P0656, Stores: S0030, S0142
Product ID: P0657, Stores: S0094, S0142
Product ID: P0658, Stores: S0030, S0094, S0142
Product ID: P0660, Stores: S0094, S0142
Product ID: P0663, Stores: S0030, S0094, S0142
Product ID: P0664, Stores: S0030, S0094, S0142
Product ID: P0665, Stores: S0030, S0094, S0142
Product ID: P0670, Stores: S0094, S0142
Product ID: P0671, Stores: S0094, S0142
Product ID: P0673, Stores: S0094, S0142
Product ID: P0674, Stores: S0094, S0142
Product ID: P0676, Stores: S0030, S0094, S0142
Product ID: P0678, Stores: S0094, S0142
Product ID: P0679, Stores: S0094, S0142
Product ID: P0680, Stores: S0030, S0094, S0142
Product ID: P0681, Stores: S0030, S0094, S0142
Product ID: P0684, Stores: S0094, S0142
Product ID: P0689, Stores: S0030, S0094, S0142
Product ID: P0690, Stores: S0094, S0142
Product ID: P0692, Stores: S0094, S0142
Product ID: P0694, Stores: S0030, S0094, S0142
Product ID: P0695, Stores: S0094, S0142
Product ID: P0698, Stores: S0094, S0142
Product ID: P0702, Stores: S0094, S0142
Product ID: P0704, Stores: S0094, S0142
Product ID: P0705, Stores: S0094, S0142
Product ID: P0707, Stores: S0094, S0142
Product ID: P0709, Stores: S0030, S0094, S0142
Product ID: P0711, Stores: S0030, S0094, S0142
Product ID: P0712, Stores: S0094, S0142
Product ID: P0714, Stores: S0094, S0142
Product ID: P0718, Stores: S0030, S0094, S0142
Product ID: P0720, Stores: S0094, S0142
Product ID: P0721, Stores: S0094, S0142
Product ID: P0724, Stores: S0094, S0142
Product ID: P0726, Stores: S0030, S0094, S0142
Product ID: P0729, Stores: S0030, S0094, S0142
Product ID: P0731, Stores: S0094, S0142
Product ID: P0733, Stores: S0094, S0142
Product ID: P0735, Stores: S0094, S0142
Product ID: P0741, Stores: S0094, S0142
Product ID: P0742, Stores: S0094, S0142
Product ID: P0747, Stores: S0030, S0094, S0142
Product ID: P0748, Stores: S0094, S0142
Number of products that are present in all 3 stores: 109
Products present in all stores:
Product ID: P0015
Product ID: P0018
Product ID: P0026
Product ID: P0035
Product ID: P0039
Product ID: P0044
Product ID: P0051
Product ID: P0055
Product ID: P0057
Product ID: P0058
Product ID: P0061
Product ID: P0062
Product ID: P0083
Product ID: P0090
Product ID: P0092
Product ID: P0099
Product ID: P0103
Product ID: P0112
Product ID: P0114
Product ID: P0116
Product ID: P0125
Product ID: P0129
Product ID: P0131
Product ID: P0132
Product ID: P0138
Product ID: P0171
Product ID: P0175
Product ID: P0182
Product ID: P0185
Product ID: P0195
Product ID: P0198
Product ID: P0201
Product ID: P0202
Product ID: P0206
Product ID: P0210
Product ID: P0238
Product ID: P0260
Product ID: P0261
Product ID: P0265
Product ID: P0269
Product ID: P0275
Product ID: P0277
Product ID: P0280
Product ID: P0286
Product ID: P0296
Product ID: P0311
Product ID: P0318
Product ID: P0325
Product ID: P0327
Product ID: P0333
Product ID: P0348
Product ID: P0353
Product ID: P0356
Product ID: P0364
Product ID: P0372
Product ID: P0374
Product ID: P0390
Product ID: P0391
Product ID: P0394
Product ID: P0408
Product ID: P0417
Product ID: P0421
Product ID: P0427
Product ID: P0428
Product ID: P0436
Product ID: P0438
Product ID: P0453
Product ID: P0455
Product ID: P0458
Product ID: P0461
Product ID: P0498
Product ID: P0499
Product ID: P0500
Product ID: P0509
Product ID: P0512
Product ID: P0525
Product ID: P0528
Product ID: P0536
Product ID: P0548
Product ID: P0549
Product ID: P0551
Product ID: P0559
Product ID: P0567
Product ID: P0569
Product ID: P0570
Product ID: P0572
Product ID: P0576
Product ID: P0590
Product ID: P0600
Product ID: P0602
Product ID: P0614
Product ID: P0625
Product ID: P0639
Product ID: P0642
Product ID: P0658
Product ID: P0663
Product ID: P0664
Product ID: P0665
Product ID: P0676
Product ID: P0680
Product ID: P0681
Product ID: P0689
Product ID: P0694
Product ID: P0709
Product ID: P0711
Product ID: P0718
Product ID: P0726
Product ID: P0729
Product ID: P0747

Sales by cluster¶

Check how many products each cluster has¶

In [ ]:
# Count the number of unique products in each cluster
product_counts_per_cluster = konya.groupby('cluster_id')['product_id'].nunique().reset_index()

# Rename the columns for clarity
product_counts_per_cluster.columns = ['cluster', 'product_count']

# Display the number of products in each cluster
print(product_counts_per_cluster)

     cluster  product_count
0  cluster_0            300
1  cluster_1              4
2  cluster_2              9
3  cluster_3             28
4  cluster_4             34
5  cluster_5             12
6  cluster_6             23
7  cluster_7             10
8  cluster_8             14
9  cluster_9             46

Check how many hierarchy1_id each cluster has¶

In [ ]:
# Count the number of unique products in each cluster
h_counts_per_cluster = konya.groupby('cluster_id')['hierarchy1_id'].nunique().reset_index()

# Rename the columns for clarity
h_counts_per_cluster.columns = ['cluster', 'hierarchy1_count']

# Display the number of products in each cluster
print(h_counts_per_cluster)

     cluster  hierarchy1_count
0  cluster_0                 4
1  cluster_1                 1
2  cluster_2                 3
3  cluster_3                 3
4  cluster_4                 3
5  cluster_5                 3
6  cluster_6                 3
7  cluster_7                 4
8  cluster_8                 3
9  cluster_9                 4

Check how many store_id each cluster has¶

In [ ]:
# Count the number of unique products in each cluster
store_counts_per_cluster = konya.groupby('cluster_id')['store_id'].nunique().reset_index()

# Rename the columns for clarity
store_counts_per_cluster.columns = ['cluster', 'store_count']

# Display the number of products in each cluster
print(store_counts_per_cluster)

     cluster  store_count
0  cluster_0            3
1  cluster_1            3
2  cluster_2            3
3  cluster_3            3
4  cluster_4            3
5  cluster_5            3
6  cluster_6            3
7  cluster_7            3
8  cluster_8            3
9  cluster_9            3

Check if there is any product that is present in more than one cluster¶

In [ ]:
# Group by product_id and count unique clusters
product_clusters = konya.groupby('product_id')['cluster_id'].nunique().reset_index()

# Filter products that are in more than one cluster
products_in_multiple_clusters = product_clusters[product_clusters['cluster_id'] > 1]

# Display products that are in more than one cluster
print("Products present in more than one cluster:")
print(products_in_multiple_clusters)

Products present in more than one cluster:
Empty DataFrame
Columns: [product_id, cluster_id]
Index: []

Check if the dimensions of the products are similiar within cluster¶

In [ ]:
# Remove duplicate entries by keeping the first occurrence
konya_unique = konya.drop_duplicates(subset=['cluster_id', 'product_id'])

# Optionally, aggregate if needed (e.g., average width if duplicates exist)
# df_unique = df.groupby(['cluster', 'product_id']).agg({'product_width': 'mean'}).reset_index()

# Display the cleaned dataset
print(konya_unique.head(10))

# Get unique clusters
clusters = konya_unique['cluster_id'].unique()

  store_id product_id       date  sales  revenue  stock  price promo_type_1  \
0    S0030      P0015 2017-01-02    0.0     0.00    4.0   2.60         PR14   
1    S0030      P0018 2017-01-02    1.0     1.81    5.0   1.95         PR14   
2    S0030      P0035 2017-01-02    2.0     4.54    1.0   2.45         PR14   
3    S0030      P0051 2017-01-02    0.0     0.00   27.0   0.70         PR14   
4    S0030      P0055 2017-01-02    0.0     0.00   12.0   3.50         PR05   
5    S0030      P0057 2017-01-02    0.0     0.00    4.0  12.90         PR14   
6    S0030      P0062 2017-01-02    0.0     0.00    5.0  19.90         PR14   
7    S0030      P0099 2017-01-02    0.0     0.00    5.0  10.90         PR14   
8    S0030      P0103 2017-01-02    0.0     0.00   13.0   2.65         PR14   
9    S0030      P0114 2017-01-02    1.0     0.42   20.0   0.45         PR14   

  promo_bin_1 promo_type_2  ... city_id_old  country_id city_code  weekday  \
0         NaN         PR03  ...        C006      Turkey     Konya      Mon   
1         NaN         PR03  ...        C006      Turkey     Konya      Mon   
2         NaN         PR03  ...        C006      Turkey     Konya      Mon   
3         NaN         PR03  ...        C006      Turkey     Konya      Mon   
4     verylow         PR03  ...        C006      Turkey     Konya      Mon   
5         NaN         PR03  ...        C006      Turkey     Konya      Mon   
6         NaN         PR03  ...        C006      Turkey     Konya      Mon   
7         NaN         PR03  ...        C006      Turkey     Konya      Mon   
8         NaN         PR03  ...        C006      Turkey     Konya      Mon   
9         NaN         PR03  ...        C006      Turkey     Konya      Mon   

   season  week holiday month_name  Year Month  
0       1     1       N        Jan  2017   Jan  
1       1     1       N        Jan  2017   Jan  
2       1     1       N        Jan  2017   Jan  
3       1     1       N        Jan  2017   Jan  
4       1     1       N        Jan  2017   Jan  
5       1     1       N        Jan  2017   Jan  
6       1     1       N        Jan  2017   Jan  
7       1     1       N        Jan  2017   Jan  
8       1     1       N        Jan  2017   Jan  
9       1     1       N        Jan  2017   Jan  

[10 rows x 34 columns]
In [ ]:
# Plot product_width for each cluster
for cluster in clusters:
    # Filter data for the current cluster
    cluster_data = konya_unique[konya_unique['cluster_id'] == cluster]

    # Create a bar plot
    plt.figure(figsize=(7, 3))
    sns.barplot(x='product_id', y='product_width', data=cluster_data, palette='viridis')

    # Customize the plot
    plt.xticks(rotation=90)  # Rotate x labels for better readability
    plt.title(f'Product Width for Each Product in Cluster {cluster}')
    plt.xlabel('Product ID')
    plt.ylabel('Product Width')
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()
In [ ]:
# Plot product_width for each cluster
for cluster in clusters:
    # Filter data for the current cluster
    cluster_data = konya_unique[konya_unique['cluster_id'] == cluster]

    # Create a bar plot
    plt.figure(figsize=(7, 3))
    sns.barplot(x='product_id', y='product_length', data=cluster_data, palette='viridis')

    # Customize the plot
    plt.xticks(rotation=90)  # Rotate x labels for better readability
    plt.title(f'Product Length for Each Product in Cluster {cluster}')
    plt.xlabel('Product ID')
    plt.ylabel('Product Length')
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()
In [ ]:
# Plot product_width for each cluster
for cluster in clusters:
    # Filter data for the current cluster
    cluster_data = konya_unique[konya_unique['cluster_id'] == cluster]

    # Create a bar plot
    plt.figure(figsize=(7, 3))
    sns.barplot(x='product_id', y='product_depth', data=cluster_data, palette='viridis')

    # Customize the plot
    plt.xticks(rotation=90)  # Rotate x labels for better readability
    plt.title(f'Product Depth for Each Product in Cluster {cluster}')
    plt.xlabel('Product ID')
    plt.ylabel('Product Depth')
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()

Time evolution¶

Check product sales evolution for each cluster¶

In [ ]:
# Ensure the date column is in datetime format
#konya['date'] = pd.to_datetime(konya['date'])

# Set the date column as the index
#konya.set_index('date', inplace=True)

# Get unique clusters
clusters = konya['cluster_id'].unique()

# Iterate through each cluster
for cluster_id in clusters:
    # Filter the dataset for the given cluster
    cluster_data = konya[konya['cluster_id'] == cluster_id]

    # Get the list of unique products within the cluster
    products = cluster_data['product_id'].unique()

    # Create a figure for the current cluster
    plt.figure(figsize=(14, 7))
    plt.title(f'Sales Over Time for All Products in Cluster {cluster_id}', fontsize=16)

    # Plot sales for each product
    for product in products:
        product_data = cluster_data[cluster_data['product_id'] == product]
        plt.plot(product_data.index, product_data['sales'], label=f'Product ID: {product}')

    # Customize the plot
    plt.xlabel('Date')
    plt.ylabel('Sales')
    plt.legend(loc='upper left', bbox_to_anchor=(1, 1), ncol=10)  # Move legend outside the plot and use 2 columns
    plt.xticks(rotation=45)  # Rotate x labels for better readability
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()

Check product stock evolution for each cluster¶

In [ ]:
# Iterate through each cluster
for cluster_id in clusters:
    # Filter the dataset for the given cluster
    cluster_data = df1[df1['cluster_id'] == cluster_id]

    # Get the list of unique products within the cluster
    products = cluster_data['product_id'].unique()

    # Create a figure for the current cluster
    plt.figure(figsize=(14, 7))
    plt.title(f'Stock Over Time for All Products in Cluster {cluster_id}', fontsize=16)

    # Plot sales for each product
    for product in products:
        product_data = cluster_data[cluster_data['product_id'] == product]
        plt.plot(product_data.index, product_data['stock'], label=f'Product ID: {product}')

    # Customize the plot
    plt.xlabel('Date')
    plt.ylabel('Stock')
    plt.legend(loc='upper left', bbox_to_anchor=(1, 1), ncol=10)  # Move legend outside the plot and use 2 columns
    plt.xticks(rotation=45)  # Rotate x labels for better readability
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()

Check product price evolution for each cluster¶

In [ ]:
# Iterate through each cluster
for cluster_id in clusters:
    # Filter the dataset for the given cluster
    cluster_data = df1[df1['cluster_id'] == cluster_id]

    # Get the list of unique products within the cluster
    products = cluster_data['product_id'].unique()

    # Create a figure for the current cluster
    plt.figure(figsize=(14, 7))
    plt.title(f'price Over Time for All Products in Cluster {cluster_id}', fontsize=16)

    # Plot sales for each product
    for product in products:
        product_data = cluster_data[cluster_data['product_id'] == product]
        plt.plot(product_data.index, product_data['price'], label=f'Product ID: {product}')

    # Customize the plot
    plt.xlabel('Date')
    plt.ylabel('Price')
    plt.legend(loc='upper left', bbox_to_anchor=(1, 1), ncol=10)  # Move legend outside the plot and use 2 columns
    plt.xticks(rotation=45)  # Rotate x labels for better readability
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()

Check product revenue evolution for each cluster¶

In [ ]:
# Iterate through each cluster
for cluster_id in clusters:
    # Filter the dataset for the given cluster
    cluster_data = df1[df1['cluster_id'] == cluster_id]

    # Get the list of unique products within the cluster
    products = cluster_data['product_id'].unique()

    # Create a figure for the current cluster
    plt.figure(figsize=(14, 7))
    plt.title(f'Revenue Over Time for All Products in Cluster {cluster_id}', fontsize=16)

    # Plot sales for each product
    for product in products:
        product_data = cluster_data[cluster_data['product_id'] == product]
        plt.plot(product_data.index, product_data['revenue'], label=f'Product ID: {product}')

    # Customize the plot
    plt.xlabel('Date')
    plt.ylabel('Revenue')
    plt.legend(loc='upper left', bbox_to_anchor=(1, 1), ncol=10)  # Move legend outside the plot and use 2 columns
    plt.xticks(rotation=45)  # Rotate x labels for better readability
    plt.tight_layout()  # Adjust layout to fit labels

    # Show the plot
    plt.show()